Limits of end-state planning

Highlights

• We asked participants to grasp and move a plastic bowl according to instruction.

• In contrast to previous studies, no strategy towards end-state comfort was found.

• End-state comfort strategies may not extend to more complex grasping movements.

Abstract

The end-state comfort effect is the tendency to use an uncomfortable initial grasp posture for object manipulation if this leads to a comfortable final posture. Many studies have replicated the end-state comfort effect across a range of tasks and conditions. However, these tasks had in common that they involved relatively simple movements, such as picking up a dowel or sliding a pan from one place to another. Here we asked whether the end-state comfort effect extends to more complex tasks. We asked participants to grasp a transparent bowl and move the bowl to an instructed location, positioning it in an instructed orientation. We either found an initial-state comfort effect or equal degrees of comfort for end-grasps and start-grasps depending on task instructions. The end-state comfort effect was not consistently observed. The results suggest that the end-state comfort effect may be restricted to relatively simple grasping movements.

Introduction

The human motor system affords a great deal of flexibility in the way movements are performed. This flexibility reflects the many degrees of freedom available to the motor system. Selecting particular movements to perform a task when many means are available, is called the degrees of freedom problem (Bernstein, 1967). The degrees of freedom problem appears to be solved in part, at least in the case of grasping movements, by favoring easy-to-control final grasps for tasks requiring great deal of final control. In those cases, people adopt awkward (extreme joint-angle) initial grasps that lead to less awkward (midrange joint-angle) final grasp postures (Rosenbaum and Jorgensen, 1992, Rosenbaum et al., 1990, Rosenbaum et al., 1996, Rosenbaum et al., 1993).

This effect was first documented by Rosenbaum et al. (1990), who asked participants to pick up a dowel placed horizontally on a pair of cradles and then to touch a target on either side of the cradles with a specified end of the dowel. Depending on which side of the dowel had to touch the target, participants grasped the dowel with an overhand or an underhand grip. Experiments showed that the choice of initial grip was determined by the comfort of the final posture. Participants chose an uncomfortable initial (underhand) grip if this led to a comfortable (thumb-up) final posture (Rosenbaum et al., 1990).

Further studies suggested that the end-state comfort effect is found in a broad range of object manipulation tasks. It was found when a dowel had to be moved to shelves placed at different heights (Short & Cauraugh, 1997), when a dowel had to be rotated (Rosenbaum & Jorgensen, 1992), when a plunger had to be moved to shelves of different heights (Cohen & Rosenbaum, 2004), and when participants were instructed to slide a pan handle in different directions (Zhang & Rosenbaum, 2008). The effect was also found in children (Adalbjornsson et al., 2008, Thibaut and Toussaint, 2010, Weigelt and Schack, 2010) and in non-human primates (Chapman et al., 2010, Weiss et al., 2007). It has also been studied in different patient groups, including those with cerebral palsy (Steenbergen et al., 2000, Steenbergen et al., 2004) or autism (van Swieten et al., 2010). End-state comfort planning has been found as well in bimanual coordination (Fischman et al., 2003, Hughes et al., 2011, Weigelt et al., 2006), in whole-body movements (Lam, McFee, Chua, & Weeks, 2006), and in passing objects from one person to another (Gonzalez et al., 2011, Herbort et al., 2012).

The end-state comfort has been taken to support the hypothesis that people plan their movements based on postures, as expressed in a computational model of movement planning (Rosenbaum et al., 1995, Rosenbaum et al., 2001). In this model, goal postures are selected from a set of stored postures and then subjected to some random variation in a search for a goal posture that best meets the task requirements. The ensuing movement is then planned as a continuous change from the initial posture to the final posture.

Finding evidence for end-state comfort planning across a broad range of tasks would provide strong support for posture-based motion planning. Pursuing this critical test, we asked whether the end-state comfort effect holds in a more complex task than has been previously studied. We asked participants to pick up a transparent plastic bowl and place it at a predefined location in a predefined orientation. This task increases the complexity of previous task in that it involves a continuum of possible grasp positions (rather than the dichotomy between underhand and overhand grasps), and it involves simultaneous translation and rotation as the bowl is moved from one place to another (translation) with its orientation changing from the first place to the second (rotation).

Whereas several studies have used a continuum of possible grasp positions (Cohen and Rosenbaum, 2004, Zhang and Rosenbaum, 2008), just one earlier study appears to have combined a continuum of grasp positions with simultaneous translation and rotation (Cohen & Rosenbaum, 2011).

If participants consistently used an end-state comfort strategy, we would expect them, by definition, to grasp the bowl in such a way that that they would adopt comfortable posture at the ends of the bowl displacements. To test this prediction, we conducted six experiments, first to replicate the findings under a range of conditions, and second, to examine possible factors involved in the effect, such as planning time and grasp flexibility. To anticipate the main result, we found evidence for initial- rather than end-state comfort in the first experiment, but then, in the subsequent experiments, we found no consistent preference either for initial- or end-state comfort. In the General Discussion section we consider the theoretical implications of this finding.